Page 1 of 2

Linear loudness vertical axis in frequency spectrum?

Posted: Fri Sep 30, 2011 6:00 pm
by zapparella
Is there a way to display vertical axis on the Spectrum on a linear scale, rather than the log scale (dB) it normally uses? (I refer to to the Power spectrum - loudness versus frequency from the Analyze tool bar - not the "sonogram" display of frequency vs time.)

The reason is to deemphasize the data at extremely low loudness (say below -30 dB).

Re: Linear loudness vertical axis in frequency spectrum?

Posted: Fri Sep 30, 2011 8:21 pm
by kozikowski
The left edge values conform to what is required. I did an analysis of a well-behaved noise signal and the top value is -36.2 and the bottom value is -39.8.

I don't think you can force it. Other elves may know better.

Koz

Re: Linear loudness vertical axis in frequency spectrum?

Posted: Sat Oct 01, 2011 12:45 pm
by steve
zapparella wrote:The reason is to deemphasize the data at extremely low loudness (say below -30 dB).
If you just want to be able to pick out the high level spikes more easily, a little trick you can try is to add a new track and generate white noise at a level of about 0.5 into that track. Now select both the original track and your noise track and view the spectrum (Analyze > Plot Spectrum).
The presence of the white noise adds a relatively high baseline level across all frequency bands, so the vertical scale is automatically adjusted .

Re: Linear loudness vertical axis in frequency spectrum?

Posted: Mon Oct 03, 2011 8:35 pm
by kozikowski
That doesn't actually cut off everything below, say -30, that compresses everything you have into a smaller display. I wonder why the poster wants that instead of just looking at the screen.

Koz

Re: Linear loudness vertical axis in frequency spectrum?

Posted: Mon Oct 03, 2011 9:01 pm
by steve
Piano music on its own - note the vertical scale range -96 dB to -23 dB
Frequency Analysis_002.png
Piano music
Frequency Analysis_002.png (41.55 KiB) Viewed 2471 times
Piano music and -6 dB white noise. Note the vertical scale range -42 dB to -23 dB.
The noise floor across the full spectrum is around -42 dB for each frequency band, so the display has re-scaled and shows the range -42 to -23 dB stretched the full vertical height rather than being squashed up in the top 1/3rd of the display.
Frequency Analysis_003.png
Piano music with white noise
Frequency Analysis_003.png (42.85 KiB) Viewed 2471 times
kozikowski wrote: I wonder why the poster wants that
So do I ;)

Re: Linear loudness vertical axis in frequency spectrum?

Posted: Tue Oct 04, 2011 4:21 pm
by zapparella
The poster (me) is a teacher teaching a course to nonscience college students in a Physics of Music course. We demonstrate in class what the spectrum of different instruments looks like, and the important features should be easily identifiable to students who don't understand logs or db scales. Why have the room noise from the air conditioning showing up as a broad peak, or a tiny (70 db weaker) resonance showing up as a peak, when these is not relevant to the main resonances of a trumpet horn?

I appreciate responders efforts to help with this. I did submit a request for a new feature, and the powers that be are considering it.

Re: Linear loudness vertical axis in frequency spectrum?

Posted: Tue Oct 04, 2011 4:37 pm
by steve
Thank you for satisfying my curiosity zapparella :)

For my own usage I would not find a linear scale to be useful.
(Considering the frequency plots in my previous posts, most of the interesting stuff is in the range of -42 dB to -24 dB, which on a linear scale is about 0.008 to 0.063, but that is probably less intuitive than the dB scale).

However, I would like the vertical axis to be controllable. When comparing two plots it is a lot easier if the scale is the same for both, but the automatic scaling prevents that.

Re: Linear loudness vertical axis in frequency spectrum?

Posted: Tue Oct 11, 2011 7:27 pm
by Gale Andrews
@zapparella: "Linear Plot Spectrum vertical axis" has now been added to Wiki Feature Requests as I promised. But it has only your vote and we are not (generally speaking) adding or improving features until after the next 2.0 major release (which is held up by fixing some project stability issues).
steve wrote:I would like the vertical axis to be controllable. When comparing two plots it is a lot easier if the scale is the same for both, but the automatic scaling prevents that.
Wiki Feature Requests has suggestions for "User-defined minimum dB level, but not via meter/waveform dB preference: (9 votes)", "Zoomable/pannable sliders (3 votes)" and "Text box input for lowest visible level (4 votes)". Would you vote for a checkbox for "automatically scale dB" or similar, so you don't always have to have the upper limit of the scale limited to the peak?


Gale

Re: Linear loudness vertical axis in frequency spectrum?

Posted: Tue Oct 11, 2011 8:35 pm
by steve
Gale Andrews wrote:Would you vote for a checkbox for "automatically scale dB" or similar, so you don't always have to have the upper limit of the scale limited to the peak?
I think that if there was an option to switch off automatic scaling then there would need to be some way for the user to set the scale. I'm not particularly concerned about "how" the user would set the scale (whether through a text box, slider, menu or whatever) but just that scale can be set manually.

Re: Linear loudness vertical axis in frequency spectrum?

Posted: Wed Oct 12, 2011 12:33 am
by Gale Andrews
steve wrote: I think that if there was an option to switch off automatic scaling then there would need to be some way for the user to set the scale. I'm not particularly concerned about "how" the user would set the scale (whether through a text box, slider, menu or whatever) but just that scale can be set manually.
I added 2 votes (including yours) for "Option to turn off dB scaling ". However the other person wanting it was happy if the resulting vertical scale was just 0 dB to the Preferences-determined minimum.


Gale